5.3.1 Introduction

This work was instigated to assess fast pyrolysis as an alternative method for tyres disposal. The primary objective was to demonstrate that short residence times and high heating rates could produce high yields of liquids from rubber. A 10-25 kg/h unit was constructed to investigate the reactor parameters of surface temperature and gas/vapour product residence time and particle size. The system is expected to be capable of using wood as a feedstock.

The process technology has been sold or licensed to Castle Capital who have installed a 1500 — 2000 kg/h plant in Halifax, Nova Scotia, using solid waste as feed. This operates at higher temperatures giving a gas product but can be modified to produce liquids. Little information is available on this plant.

There are three types of product — solid, liquid and gas. Each type and a selection of examples is given below. For each individual product, there are three sets of data that can be provided as indicated under 1,2 and 3 below.

Gas LHV fuel gas, MHV fuel gas, SNG, hydrogen, ammonia,

Liquid Naphtha, diesel, gasoline, fuel oil, bio-oil, liquefaction oil, tar, methanol, ethanol, fuel alcohol, MTBE, ETBE, char-water slurry, char — oil slurry,

Solid Charcoal.

In addition chemicals might be included as a separate category as follows:

Chemicals Methanol, ethanol, ammonia, benzene, toluene, xylene, phenol, hydroxyacetaldehyde, acetic acid, and several hundred other specialities such as hydroxyacetaldehyde, levoglucosan etc.

a Physical and chemical properties

For each product, basic data on physical and chemical properties should be provided. This can range from simple to extensive according to resources available and might include:

• Higher heating value — MJ/kg,

• Lower heating value — MJ/kg,

• Elemental analysis — weight basis — CHONS etc. and minor and trace components,

• Molecular weight,

• Viscosity,

• Density,

• Solubility in water, hydrocarbons, other solvents,

• Flash point,

• Explosive limits,

• Toxicity,

• etc.

b Production methods

The sources by technology can be identified for each product and cross referenced to the TECHNOLOGY file.

c Production costs

It would be possible to derive simple relationships that express product cost as a function of plant size and feed cost. This could not be achieved in the current budget and time scale.

1.1 INTRODUCTION

The potential offered by biomass and solid wastes for solving some of the world’s energy and environmental problems is widely recognised. The energy in biomass may be realised either by direct use as in combustion to give heat, or by conversion and upgrading into a more valuable and usable fuel such as fuel gas or fuel oil or higher value products for the chemical industry. Liquid products have significant advantages in handling, storage, transport and substitution for conventional fuels and pyrolysis is being rapidly developed for direct production of both crude liquids for direct fuel oil substitution and production of hydrocarbons for more technically demanding applications and transport fuels. There is a further advantage in electricity generation of being able to de-couple fuel production from electricity generation through fuel storage which is not possible in gasification or combustion systems.

Biomass has received considerable attention as a renewable energy resource after the oil crises of the last 20 years. Pyrolysis in particular has been researched and developed for the economic production of fuel products that may be readily integrated into the energy infrastructures of both industrialised and developing countries. More recently, attention has focussed onto much higher value chemicals either as unique specialities or as substitutes for petroleum derived products.

Particulate levels may be high from char and ash carry-over. Separation of solids and liquids is poorly understood with reliance placed on primary separation in the vapour phase downstream of the reactor before condensation. Efficient separation inevitably causes some condensation or precipitation and carefuf design is essential. Solid separation in the liquid phase is not believed to have been studied, but is very likely to be troublesome. However, it is clear that a fairly high level of charcoal can be assimilated in the liquid product, for example Alten reported up to 15% (126) although some lumpiness was evident in the bio-oil. Both char particle size and proportions will influence the liquid product quality. This is why research into char-oil mixtures could prove valuable.

3.6.3 Oxygen content

The oxygen content of the pyrolysis liquid is very high, at up to 40% wt. When produced from dry or low moisture content feeds it typically has a heating value a little below that of the biomass feed in the range 16-20 MJ/kg depending on moisture content (around 22-23 MJ/kg on a dry basis). The oxygen content arises from the wide range of oxygenated compounds including phenols and polyphenols, which are one of the chemical products that can be recovered as a valuable fraction (119).

5.16.1 Summary

The objectives of the work are to develop and operate a novel ablative pyrolysis reactor for the production of liquids in high yields form wood feedstock under a range of operating parameters, primarily reactor temperature and residence time. The process concept has been successfully developed at 3 kg/h and oil yields of up to 80 % wt. on dry feed have been obtained. A scaled up version is being developed.

5.16.2 Description

Figure 5.14 Aston University Ablative Pyrolysis Reactor:

Simplified Diagram.

Dried biomass particles of up to 6.35 mm are fed into the nitrogen purged reactor from a sealed screw feeder. Four asymmetric blades rotating at speed up to 200 rpm generate a mechanical pressure on the particles, pressing the particles onto the heated reactor base, typically heated to 600°C. The mechanical action of the blades causes the particles to pyrolyse (thermally erode) under the conditions of high relative motion to the heated reactor surface. The product vapours and gases are removed from the reactor with the diluting nitrogen under the action of a vacuum pump.

The primary liquids are recovered in an ice cooled condenser arrangement with final stable product vapours in the non-condensable gases being removed by a cotton wool filter. The remaining gas is dried with a molecular sieve before discharge from the vacuum pump to the gas analysers and the fume hoods.

5.16.3 Products

Table 5.14 gives some results from early experiments.

This work is continuing with new heat transfer relationships being derived and further experiments carried out at a range of reactor temperatures and residence times. A scaled up version of the reactor is being developed. There is associated work on catalytic pyrolysis, chemicals recovery from bio-oil and physical property measurement.

Secondary pyrolysis kinetics has been studied by number of researchers to account for the conversion of primary liquids to secondary products such as char, tar and gases. Antal (45, 46) proposed the conversion of primary vapours to either gases or tars by a temperature based competitive reaction scheme (Figure 2.8).

Primary Pyrolysis Secondary Pyrolysis

The first reaction produced more permanent gases by cracking the reactive volatile matter to smaller, less reactive species. The second reaction produced refractory condensable materials, which may be tar or some combination of water-soluble organic compounds.

where:

Cb the mass fraction of carbon in biomass

Cv the mass fraction of carbon composing the reactive volatile matter (carbon in volatile matter/carbon In sample pyrolysate)

Cgi, Cg2 the mass fraction of carbon composing the permanent gases

C the mass fraction of carbon composing the refractory condensable materials, including the tars.

Recently kinetic models for the secondary decomposition of primary pyrolysis tars have been proposed by Liden (80), Diebold (56) and Knight et al. (81). Liden (80) and Diebold (56) using different reactor configurations, proposed similar kinetic models. The reaction scheme used is shown in Figure 2.9.

The kinetic expression used for the estimation of the yield of liquid products and the values of the kinetic parameters used for each model are listed below:

1-exp(-k3)> k3 q j

where

k3: the reaction rate constant for the oil decomposition step [s-1]

q: the mean residence time for the oil decomposition [s]

x0: the theoretical ‘ultimate’ oil yield

with: k3 = 4.28 x 106 exp ^ ^ j s-1; x0 = 0.703

and; k3 = 1.55 x 1 o5 exp f’8^.34) s-1; x0 = 0.78 or 0.76

for Liden and Diebold respectively.

The reaction scheme used by Knight et al. (81) is shown below in Figure 2.10 with the following reaction rate expression :

k1b5

(k2-ki )[exp (-k2t)]

with

kj = 1.483 x 106 exp ^ s-1 k2 = 23.12 exp s_^ b§ = 0.811

where,

ki: reaction rate constant for the first order production of oil

k2: reaction rate constant for the first order decomposition of oil

b5: maximum fractional conversion of wood to oil

ki

W00d——————— ► Gas + Oil + Char

4%

Gas + Char

Figure 2.10 The Reaction Scheme of Knight et al. (81)

VasaJos et al. (82) and Scott et al. (83) tested the above models using their own experimental data. Vasalos et al. (82) found that using Liden’s parameters, they obtained a better fit of the liquid yields of -20% for the particle size range 300-425 mm, while Diebold’s parameters gave a better fit of ± 10% for the particle size range 500-600 mm. Knight’s et al. model did not predict satisfactorily liquid yields for either particle size range. They concluded that the variations between the predicted values and the experimental results could be attributed to the exclusion of the water yield in the reaction mechanism, the residence times used, and the type and size of biomass tested.

Scott and Piskorz (83) found when testing the models of both Liden and Diebold, that the predicted liquid yields agreed with achieved yields within ± 10% for the temperature range 500-700°C with residence times of up to 1 second. Low predictions of liquid yields at the highest temperature were attributed to the assumed water yield, the constant x0 parameter or the total yield being normalised

to 100%.

2.3.4.1 Summary

As pyrolysis is a very complex process and the different intermediates formed are difficult to collect and identify, various approaches have been used to develop kinetic models. Most predict weight loss rather than product yield and distribution. The kinetic parameters vary from one model to another because they are very sensitive to experimental conditions. One research group found that even a decrease of 1 kCal/mol (from 31.8 to 30.8) in the activation energy of tars caused the predicted value of the liquid yield to increase by approximately 16 % (84).

Stagewise models have been discussed earlier with regards to pyrolysis pathways and mechanisms.

Types of pyrolysis reactors and the methods of heating employed are shown in Table 4.2. Configurations are basically similar to those employed in gasification although a wider range of unusual combinations and configurations have been devised. Key features of pyrolysis processes are the method of heating which was described and discussed in Chapter 3.

The organisations currently involved in fast pyrolysis for production of primary liquids are listed alphabetically in Table 4.3 with an indication of the maximum capacity achieved to date. Each is summarised in the sections below. Figure 4.1 shows how the maximum size of installed plant has grown steadily since pyrolysis was discovered around 1979, and forecasts major plant installations by 2000.

1980 1985 1990 1995 2000

Date

Direct production of speciality chemicals by concurrent catalytic processing with pyrolysis has also attracted some attention. There has been scattered research on the effects of adding various catalysts to biomass prior to pyrolysis. Addition of sodium chloride increased yields of char and of certain chemicals such as glycolaldehyde (hydroxyacetaldehyde), and suppressed levoglucosan production (41). Similar effects were found with zinc chloride, except that higher yields of furfural were obtained (50), and also with cobalt chloride additions to almond kernels which gave higher yields of 2-fura! dehyde (29).

Pyrolysis in molten salts gave significant yields of acetone and hydrocarbon gases (51) and under different conditions, gave high yields of relatively pure hydrogen (above 90% vol. with the balance methane) (52, 53, 54). It was postulated that careful temperature control and temperature variation such as ramping or pre-

dissolution at low temperature could lead to higher yields of potentially valuable speciality chemicals. It was not clear if there were any catalytic effects present, or if the results were derived from physical absorption of carbon dioxide in the alkaline melt with possible effects on the equilibrium through the shift reaction as demonstrated by Hallen (55). The economic and energetic consequences of melt regeneration have not been evaluated, but are recognised as significant (51).

Hydrogen has been used as a reactive gas in a catalyticaily modified atmospheric fluid bed flash pyrolysis using nickel process to give 70-75% conversion to a gas containing 85-90% methane (48, 56, 57), and olefins have also been produced in interesting yields (57). Although this is referred to as hydropyrolysis, this term is more usually employed for high pressure systems. This route has not yet been used to derive liquids although there are many interesting possibilities based on pressure processing which may be viable at the lower temperatures that are optimal for liquids production as well as atmospheric pressure catalytic pyrolysis using, for example, modified zeolites which are described below.

Integrated catalysis and flash pyrolysis has been carried out on lignins for improved cracking to fuels and chemicals over a temperature range of 500 — 800°C (58).

A review of recent developments in thermal and thermo-catalytic biomass conversion has been published (3) and the opportunities for chemicals production reviewed (59).

In ail cases of chemicals recovery, there is little evidence that the higher yields of specific chemicals will be economically viable or that there are markets for the products. Addition of environmentally sensitive materials such as chloride may cause treatment and disposal problems for wastes and by-products. For char production, addition of chloride can only be deleterious and expensive in most applications for the charcoal.

Some char is inevitably carried over from cyclones and collects in the liquid. Subsequent separation has proved difficult. Some success has been achieved with hot gas filtration in a ceramic cloth bag house filter (11) and also candle filters for short run durations. Liquid filtration has also proved difficult as the liquid can have a gel-like consistency, apparently due to some interaction of the lignin — derived fraction with the char.

This aspect of char reduction and/or removal will be increasingly important as more demanding applications are introduced which require lower char tolerances in terms of particle size and total quantity. Possible solutions include changing process conditions to reduce the nature of the pyrolytic lignin, increasing the

degree of depolymerisation of the lignin-derived fraction of the liquid, changing the feedstock to one with a lower lignin content, or adding chemicals to the liquid for example to improve handling properties or reduce char-lignin interactions.

It must not be forgotten that an alternative solution is to modify the application to accept a high char content bio-fuel-oil.

3.3.8 Ash separation

The alkali metals from biomass ash are present in the char in relatively high concentrations and cannot be readily separated except by hot gas filtration which is undergoing development.

The objectives of the work were to produce pyrolysis vapours for catalytic treatment as a parallel study to the ablative pyrolysis work being performed at NREL using the same principle of contacting biomass with a heated surface under conditions of high relative motion and applied pressure. The work was fundamental in assessing the requirements for ablation and preliminary yield data was obtained. The work is significant in establishing some of the basic work on ablative pyrolysis.

5.4.2 Description

Reed and Cowdrey constructed a "heat flux concentrator" initially to investigate ablative pyrolysis comprised of a drill press which forced and rotated wood dowels into a 1.2 cm diameter tapered hole in a heated copper block (6). The vapours produced emerged through 12 holes in the bottom of the block and were condensed and collected in traps and a gas burette. The temperature range used was 500-700°C and total liquid yields of over 50% on an as fed basis were obtained. The forced contact of the wood with the concentrator however caused the holes to be plugged after about 10g of wood was fed.

A "pyrolysis mill" was then constructed as shown in Figure 5.4 (7,8,9,10,11). This second reactor was designed using the principles of a conventional mill for grain (12). The pyrolysis mill consists of a stationary upper "stone" and a rotating lower "stone" which are both of copper and in contact with heaters. The lower plate is rotated at speeds up to 80 rpm, the pressure being controlled with a spring. The reactor wall is heated to 300-400°C to prevent pyrolysis vapour condensation inside the reactor. Wood particles enter the reactor through the upper plate and the vapours escape through the plate gap and then to a series of four liquid traps. Char and ash accumulate inside the reactor. Reed estimated the average heat flux

in the first reactor to be 6.8 W/cm^ and 5.08 W/cm^ in the second reactor (13).

/

Some results for the contact pyrolysis mill are presented in Table 5.6. Using this reactor system, total liquid yields of up to 54 % of liquid based on dry feed have been achieved. The only feedstock tested was bone dry wood. Feed rates of up to 0.2 kg/h were achieved with run times up to 1.5 hours.

One of the main problems was the escape of the pyrolysis vapours through the plates due to plugging of the holes in the disc. Unreacted feed and char remains in the hot reactor environment where it undergoes more conventional pyrolysis. Scale up of this system could be difficult. No vapour residence time values are quoted. Cowdrey did not take into account the variation of the wood decomposition temperature with heat flux and reactor heated surface temperature as discussed by Lede in his design study (14). The concept is interesting, but the problem with using disks is that variable particle size causes processing problems due to disk spacing. Small particles will, therefore, not be under applied pressure and will more slowly carbonise on the hot plates. Relative motion between particles and disk cannot be maintained as the particles tend to stick to one surface and not move as intended. Ribbed disks did not solve the problem. No further work has been carried out or is planned and the reactor has been dismantled (15).